Advances in Engineering Research (AER), volume 117 2nd Annual International Conference on Electronics, Electrical Engineering and Information Science (EEEIS 2016) Educational data mining for decision-making: a framework based on student development theory Xiao-Feng Lei1,2 1School of Ideological and Political Theory, Beihang University Beijing, 100191, China 2Graduate School of Education, Peking University Beijing, 100871, China E-mail: [email protected] Ming Yang1 1School of Ideological and Political Theory, Beihang University Beijing, 100191, China E-mail:[email protected] Yi Cai†,3 3College of Information Science and Technology, Beijing University of Chemical Technology, Beijing, 100029, China †E-mail: [email protected] Applying Educational Data Mining (EDM) for decision making is an emerging interdisciplinary research field. From the view of student development, this paper presents a framework for educational decision making, which can explore some laws and characteristics in student development to improve educational decision and quality. The paper also presents a case study to verify the effectiveness of the framework. Keywords: educational data mining; decision making; student development theory. 1. Instruction At present, using data to make decisions is not new [1]. It often happens in the business domains as well as the educational fields. For educational institutions, one of the biggest challenges is the exponential growth of educational data and the use of this data to improve the quality of managerial decisions [2]. As an increasingly emerging interdisciplinary area, Educational Data Mining (EDM) is concerned with developing methods to explore the unique types of data that come from educational environments [3]. One critical and prominent objective of data mining in education is to enable data-driven decision-making for Copyright © 2017, the Authors. Published by Atlantis Press. 628 This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). Advances in Engineering Research (AER), volume 117 improving current educational practice and learning materials, and then for serving students’ learning and boosting educational quality. In educational environment, higher education including universities or colleges will find larger and wider applications for data mining than its counterpart in other educational sectors because higher educational institutions carry at least three duties that are data mining intensive: 1) teaching that concerns with the transmission of knowledge, 2) scientific research that relates to the creation of knowledge, and 3) institutional research that pertains to the use of knowledge for decision making [4]. It means that higher educational institutes seek more efficient technology to better manage and support decision making procedures or assist them to set new strategies and plan for a better management of the current processes [5]. In previous studies, data mining technology has been applied in higher education fields, such as improving students’ academic performance [6][7], selecting courses, measuring their retention rate [8], and managing the grant fund of an institution. However, most of these studies mainly focus on the data of the educational processes or didactical contents, and few of them regard the data organized according to educational objects as the research view. These objects usually comprise student, academic faculty, administrative staff, and alumni. For example, if student is regarded as the research object, the term of student’s development from enrollment to graduation in a university can be observed. Furthermore, the data of this term can be stored and organized based on student development theory to data mining for decision-making. It will benefit to explore some new laws and characteristics in student development to improve educational decision and quality. In order to understand the progress that how data mining works for decision making from student’s view, this paper presents a research framework that descripts an idea based on Student Development Theory Model. The paper is structured as follows. Section 2 presents the review of EDM including its concept, process, application methods and tools. Section 3 presents how EDM applies for decision-making. Section 4 presents the improved model of Student Development Theory. Section 5 presents a framework for decision-making, and Section 6 gives a case study of managerial decision making on the development of graduate students. Section 7 presents the conclusion and future work. 2. Educational data mining (EDM) Since data mining technique has been applied in many domains such as business and biology, it seems not necessary to regard education as a specific research sector. However, educational data has several typical characteristics, such as multiple hierarchies, complexity in context, interaction among objects, and 629 Advances in Engineering Research (AER), volume 117 transformation along with time. These data can be generated by any type of information system supporting education and administration. For instance, for a university, these data can include the background data of students, didactical data throughout a course or an academic quarter, daily data of recreation or sport activities, and so on. These characteristics lead to further consideration and research on the application of data mining in educational field. As an interdisciplinary area, EDM is associated with at least three disciplines: computer science, education, and statistics. Conceptually, it is also closely related with knowledge discovery in databases (KDD), computer-based education, machine learning, and learning analytics (LA). According to the definition of the website of International EDM Society, EDM is an emerging discipline, concerned with developing methods for exploring the unique and increasingly large-scale data that come from educational settings, and using those methods to better understand students, and the settings in which they learn [9]. In short, it is the application of data mining techniques in educational environment as a specific knowledge discovery. Fig. 1. Application Process of Educational Data Ming Different from that of the general data mining, the application process of EDM can be seen as an iterative cycle of hypothesis formation, testing, and refinement (see Fig. 1) [3]. In this process, hypothesis formation and refinement present the educational views in particular. In fact, due to the presence of hypothesis formation and refinement, the iterative cycle of EDM looks like growing up to a robust educational system, which is not only to realize knowledge discovery, but also to feedback useful information of decision- making for better educational environment. Hypothesis formation defines the educational object of data mining differing from other domains, and provides a prerequisite hypothesis model based on educational theories. Therefore, the research object of EDM is not a disordered dataset but a meaningful dataset of the educational entity. In this paper, Student Development Model is presented as the hypothesis formation. Refinement is feedback to propose educational decisions to improve or optimize the educational environment. It means that refinement involves some matters of decision-making methods to make EDM more efficient and accurate. 630 Advances in Engineering Research (AER), volume 117 At present, there are a number of popular methods within EDM. Some of them are widely acknowledged to be universal across types of data mining, such as prediction, clustering, relationship mining, outlier detecting, social network association (SNA), process mining, and text mining. Based on these popular methods, a lot of general free and commercial DM tools and frameworks emerged to be used for mining datasets from any domain or research area [10], such as Weka, SAS Enterprise Miner, SPSS Clementine, IBM Intelligent Miner, and DB Miner. In Romero’s study [3], some other methods, particularly for EDM including the distillation of data for human judgment, discovery with models, knowledge tracing (KT) and nonnegative matrix factorization, were concluded. Correspondingly, there are some tools to specify these methods to solve different educational problems. In the case study of this paper, data mining is run by Weka because the data is formed with Student Development Model but not complex. 3. Educational Data Mining for Decision Making As a thought behavior of human being, the process of decision making generally includes the following procedures: problem finding, goal settlement, matter materials analysis, decision formation and implement. Each step content of this process may differ from various domains that demonstrate essential differences in natural laws or rules. Such as in educational fields, the process presents the characteristics of decision according to the educational phenomena and laws. In educational areas, there are large amounts of educational materials to be collected for decision making, such as administrative documents, course syllabus, academic scores, and record data of students’ activities. Among these materials, educational data has been applied to decision making, which is also called data-driven decision making (DDDM) in education [11]. Data may be used for several purposes according to the different demand levels that influence the nature of the decision making process. Teachers can use assessment data to reflect on their own teaching practice, to create